A Battlefield Situation Awareness Management Method and System Based on Panoramic Vision

Abstract

The present application relates to the field of situation awareness, and more particularly to a battlefield situation awareness management method and system based on panoramic vision. The method comprises the following steps: S1: acquiring panoramic vision data and preprocessing, and then classifying the field of view range; S2: constructing a range scaling model according to the field of view range classification result, and obtaining battlefield situation prediction factors by using the range scaling model; S3: classifying the battlefield situation safety level according to the battlefield situation prediction factors, and determining the visual display content according to the safety level classification result; S4: forming a closed loop system by the above steps. Through panoramic vision data preprocessing and field of view range classification, the present application ensures data quality; by constructing a range scaling model, the present application improves situation awareness accuracy; by dynamically adjusting the buffer area, the present application optimizes the marching strategy; by a closed loop system for continuous optimization, the present application ensures long-term stability and reliability, and significantly improves battlefield situation awareness efficiency and effect.

Description

Technical Field

[0001] This invention relates to the field of situational awareness, and in particular to a battlefield situational awareness management method and system based on panoramic vision. Background Technology

[0002] With the increasing complexity and uncertainty of modern warfare, battlefield situational awareness has become crucial in determining victory or defeat. Traditional battlefield situational awareness methods primarily rely on single sensors or cameras with limited field of view, making it difficult to acquire comprehensive and real-time information, thus limiting the accuracy and timeliness of decision support. Specifically, traditional methods suffer from the following main problems:

[0003] 1. Incomplete information acquisition: Traditional sensors and cameras can usually only cover a limited field of view and cannot achieve 360-degree panoramic field of view coverage, making it easy to miss key information.

[0004] 2. Insufficient real-time performance: Traditional methods have slow data processing and transmission speeds, making it impossible to achieve real-time updates and affecting the timeliness of decision-making.

[0005] 3. Limited data processing capabilities: Traditional equipment has limited computing power, making it difficult to process large amounts of complex data, resulting in low information processing efficiency.

[0006] 4. Poor system flexibility: Traditional systems are complex to build and maintain, require professional technicians, and are difficult to troubleshoot, which affects rapid deployment and maintenance.

[0007] To address these issues, panoramic vision technology emerged. Using multi-view cameras and fisheye lenses, panoramic vision technology achieves 360-degree panoramic field of view coverage, providing all-around, blind-spot-free battlefield information. However, panoramic vision technology also has some disadvantages:

[0008] 1. Huge amount of data: The generated image data is huge, which places high demands on storage and transmission, affecting decision-making efficiency.

[0009] 2. High computing power requirements: Image stitching, correction and real-time processing require powerful computing capabilities. Edge devices need to have high-performance processors and sufficient memory, otherwise it will increase the image processing latency.

[0010] 3. System complexity: The system is complex to build and maintain, requires professional technical personnel, and troubleshooting is difficult, which affects rapid deployment and maintenance.

[0011] 4. Poor environmental adaptability: Extreme conditions (such as high temperature, low temperature or strong electromagnetic interference) affect performance, causing instability of electronic equipment and affecting the normal operation of the system.

[0012] Therefore, to overcome the limitations of traditional methods in information acquisition, processing, and transmission, this study proposes a battlefield situational awareness management method and system based on panoramic vision. This system aims to provide more comprehensive, real-time, and accurate battlefield situational information by optimizing data processing flows, enhancing computing power, and improving system flexibility and adaptability, thereby improving the efficiency and accuracy of decision support. Summary of the Invention

[0013] To overcome the shortcomings of traditional methods in information acquisition, processing, and transmission, this invention provides a battlefield situation awareness management method and system based on panoramic vision.

[0014] The technical implementation of this invention is: a battlefield situation awareness management method based on panoramic vision, comprising the following steps:

[0015] S1: Acquire panoramic visual data and preprocess it, then classify the field of view.

[0016] S2: Based on the field of view classification results, construct a range scaling model and use the range scaling model to obtain battlefield situation prediction factors;

[0017] S3: Classify the battlefield situation security level based on battlefield situation prediction factors, and determine the visualization display content based on the security level classification results;

[0018] S4: Form the above steps into a closed-loop system and optimize the system by continuously collecting new data and feedback.

[0019] Preferably, the step of acquiring panoramic visual field range data and preprocessing it, and then classifying the field range, includes: the preprocessing includes data cleaning and standardization; the classification of the field range includes dividing the field range into front view area, rear view area, side view area, top view area, and bottom view area.

[0020] Among them, the angles of the front view area, rear view area, side view area, and upper view area are the first observation angles, and the angle of the bottom view area is the second observation angle.

[0021] Preferably, the step of constructing a range scaling model based on the field of view classification results and using the range scaling model to obtain battlefield situation prediction factors includes: constructing a scaling range model using the forward, rear, and side view data from the first observation angle; obtaining a training set of the forward, rear, and side view data; inputting the training set data into the scaling range model to obtain a trained scaling range model; and using the trained scaling range model to obtain battlefield situation prediction factors.

[0022] Preferably, obtaining the trained scaling range model includes: constructing a fan-shaped region based on the field of view and the farthest observation distance as the radius; dividing the fan-shaped region into a first region and a second region, wherein the side closer to the visual sensor is the first region and the side farther from the visual sensor is the second region;

[0023] The specific division is based on the following criteria: First, a comprehensive value is obtained based on computational complexity, computational data volume, and computational resource requirements. Then, the boundary between the first and second regions is determined by combining the SVM classification model. Finally, the area of ​​each inefficient region in the second region that does not meet the requirements of real-time battlefield is calculated using a formula.

[0024] Preferably, the process involves first obtaining a comprehensive value based on computational complexity, computational data volume, and computational resource requirements; then determining the boundary between the first and second regions using an SVM classification model; and finally calculating the area of ​​each inefficient region in the second region that does not meet the requirements of real-time battlefield using a formula. The comprehensive value calculation formula is as follows:

[0025]

[0026] in, This is a composite value. Weights for calculating complexity, To calculate the weight of resources, To calculate the weight of the data volume, To calculate the normalized value of the complexity, To calculate the normalized value of the resource, To calculate the normalized value of the data volume;

[0027] The formula for calculating the total area of ​​the second region is as follows.

[0028]

[0029] in, This represents the total area of ​​the second region. The field of view is in radians. The furthest observation distance, The radius of the first region;

[0030] The area calculation formulas for each inefficient area in the second region are as follows:

[0031]

[0032] in, For the ( , The area of ​​) inefficient areas, The angular increment for the area of ​​each inefficient region, in radians. The radius increment for the area of ​​each inefficient region. The radius of the first region is... This is a radius index for the area of ​​inefficient regions. An angular index for the area of ​​inefficient regions;

[0033] The normalization calculation formula is as follows:

[0034]

[0035] in, The original value of a certain feature. This is the minimum value of the feature. The maximum value of this feature. The value is the normalized value, ranging from [0, 1].

[0036] The core calculation formula of the SVM classification model is as follows:

[0037]

[0038] in, For the weight vector, For the input feature vector, This is a bias term.

[0039] Preferably, the step of classifying the battlefield situation security level according to the battlefield situation prediction factor and determining the visualization display content according to the security level classification result includes: obtaining the area of ​​each inefficient area in the second region; constructing a buffer region at the junction of the first region and the second region, with the two endpoints of the buffer region located on the boundary lines of the first region and the second region respectively; constructing an angle between the two endpoints of the buffer region and the center line between the two inefficient areas to form an angled region; determining the curvature of the vertex position of the angled region, using the change in curvature as the data source for the prediction factor, and using the prediction factor to classify the battlefield situation security level; the buffer region is defined as a dynamic region set at the junction of the first region and the second region to verify and confirm the reliability of the panoramic visual data;

[0040] The specific division of the region is based on: obtaining the starting point of the area of ​​each inefficient region in the second region, connecting the two ends of the buffer region to the starting points of the two inefficient regions respectively, forming two lines; if the intersection of the two lines is located at the farthest end of the second region, then the region where the intersection is located is determined as the angle construction region.

[0041] Preferably, determining the curvature at this location includes: the formula for calculating the curvature is as follows,

[0042]

[0043] in, Let f(x) be the curvature, and f(x) be a function of the curve. Let f(x) be the first derivative. It is the second derivative of f(x).

[0044] Preferably, the construction of predictive factors for classifying battlefield situational safety levels includes: recording the curvature calculation values ​​at each location and recording the curvature changes at each location over time, with a recording time interval of t seconds; extracting the curvature changes over time for the forward-looking area, side-looking area, and rear-looking area from the recorded data; using the curvature changes over time in each orientation as a predictive factor sequence; using an LSTM model with the predictive factor sequence as input to predict the optimal route for the next movement; using the predictive factor sequence to generate a recommended route for movement; using the difference between the optimal route and the recommended route as the criterion for safety level evaluation; and using an RNN model with the difference value as input to evaluate the safety level.

[0045] The method of using the difference between the optimal route and the recommended route as the criterion for judging the safety level includes: acquiring second observation angle data to supplement and verify the difference between the optimal route and the recommended route; if traveling along the optimal route, verifying the rationality of the optimal route using the second observation angle data; if the second observation angle data meets the standard of the optimal route, increasing the recommended position value of the recommended route; if the second observation angle data does not meet the standard of the optimal route, decreasing the recommended position value of the recommended route.

[0046] Preferably, the above steps constitute a closed-loop system, and the system is optimized by continuously collecting new data and feedback, including: obtaining the difference between the optimal route and the recommended route, and using the difference result as the scaling range in the range scaling model, i.e., the buffer area; the buffer area is defined as a dynamic area set around the travel path for adjusting and optimizing the travel strategy.

[0047] If the difference between the optimal route and the recommended route is negative, the buffer zone moves closer to the first region by shifting its boundary k meters towards the first region. If the difference is positive, the buffer zone moves closer to the second region by shifting its boundary k meters towards the second region. If the difference is zero, the second region expands outward according to the Transformer model by widening its boundary. These steps form a closed-loop system. By continuously collecting new data and feedback, the system is optimized to ensure continuous improvement and performance enhancement.

[0048] A battlefield situation awareness and management system based on panoramic vision includes:

[0049] Data acquisition and preprocessing module: used to acquire panoramic visual data and perform preprocessing;

[0050] Field of view classification module: used to classify the field of view;

[0051] Range scaling model module: Used to build range scaling models and obtain battlefield situation prediction factors;

[0052] Security Level Classification and Visualization Module: Used to classify the security level of the battlefield situation based on battlefield situation prediction factors and determine the content to be displayed in the visualization.

[0053] Feedback optimization module: Used to optimize the system by continuously collecting new data and feedback;

[0054] Communication and storage module: responsible for data transmission and storage.

[0055] Beneficial effects: First, preprocessing and field-of-view classification of panoramic visual data ensured data quality and consistency. Second, the construction of a range scaling model, utilizing battlefield situation prediction factors, improved the accuracy of situational awareness. Next, security level classification and visualization provided intuitive decision support. Furthermore, dynamically adjusting buffer zones and flexibly optimizing movement strategies based on the difference between the optimal and recommended routes enhanced the system's adaptability and security. Finally, the construction of a closed-loop system continuously collected new data and feedback, enabling continuous system optimization and performance improvement, ensuring long-term stability and reliability. These measures collectively improved the efficiency and effectiveness of battlefield situational awareness, providing strong support for decision-makers. Attached Figure Description

[0056] Figure 1 This is a flowchart of a battlefield situation awareness management method and system based on panoramic vision according to the present invention;

[0057] Figure 2 This is a schematic diagram of the structure of a battlefield situation awareness management system based on panoramic vision according to the present invention. Detailed Implementation

[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] Example 1: A battlefield situational awareness management method based on panoramic vision, such as Figure 1 As shown, it includes the following steps:

[0060] S1: Acquire panoramic visual data and preprocess it, then classify the field of view.

[0061] S2: Based on the field of view classification results, construct a range scaling model and use the range scaling model to obtain battlefield situation prediction factors;

[0062] S3: Classify the battlefield situation security level based on battlefield situation prediction factors, and determine the visualization display content based on the security level classification results;

[0063] S4: Form the above steps into a closed-loop system and optimize the system by continuously collecting new data and feedback.

[0064] The process involves acquiring and preprocessing panoramic visual field range data, and then classifying the field range, including: the preprocessing includes data cleaning and standardization; the classification of the field range includes dividing the field range into front view area, rear view area, side view area, top view area, and bottom view area.

[0065] Among them, the angles of the front view area, rear view area, side view area, and upper view area are the first observation angles, and the angle of the bottom view area is the second observation angle.

[0066] To further explain, the field of view, through clear classification and processing steps, manages and analyzes data more effectively, improving the overall performance and accuracy of the system. The field of view is divided into three sections: Forward View: The field of view in front of the tank, primarily used to detect enemy activity and obstacles ahead. Rear View: The field of view behind the tank, primarily used to monitor enemy activity and friendly forces behind it. Side View: The field of view on both sides of the tank, primarily used to monitor the safety of the flanks. Upper View: The field of view above the tank, primarily used to detect aerial threats, such as drones and helicopters. Lower View: The field of view below the tank, primarily used to detect ground obstacles and potential traps. Through clear classification and processing steps, data is managed and analyzed more effectively, improving the overall performance and accuracy of the system. Through the above-described field of view classification and data processing flow, the system can identify and classify targets more quickly, providing real-time battlefield situational information and helping tank commanders make more informed decisions; the lower view is classified separately because it is primarily used to detect ground obstacles and potential traps.

[0067] Based on the field of view classification results, a range scaling model is constructed, and battlefield situation prediction factors are obtained using the range scaling model. This includes: constructing a scaling range model using the forward, rear, and side view data from the first observation angle; obtaining a training set of the forward, rear, and side view data; inputting the training set data into the scaling range model to obtain a trained scaling range model; and using the trained scaling range model to obtain battlefield situation prediction factors.

[0068] To further explain, the model is built using data from the first observation angle (forward, rear, and side views) to ensure its relevance and effectiveness. By using data from specific regions to build and train the model, it better captures and reflects the characteristics of the battlefield situation, improving the model's accuracy and reliability. Training data from the forward, rear, and side views ensures the model can handle situational information from different directions, improving its generalization ability. The trained, scalable model is used to extract battlefield situational prediction factors, providing support for subsequent situational analysis and decision-making. Again, by using data from specific regions to build and train the model, it better captures and reflects the characteristics of the battlefield situation, improving the model's accuracy and reliability.

[0069] Obtaining a trained scaling range model includes: constructing a fan-shaped region based on the field of view and the farthest observation distance as the radius; dividing the fan-shaped region into a first region and a second region, where the side closer to the visual sensor is the first region and the side farther from the visual sensor is the second region;

[0070] The specific division is based on the following criteria: First, a comprehensive value is obtained based on computational complexity, computational data volume, and computational resource requirements. Then, the boundary between the first and second regions is determined by combining the SVM classification model. Finally, the area of ​​each inefficient region in the second region that does not meet the requirements of real-time battlefield is calculated using a formula.

[0071] Further explanation involves constructing a fan-shaped region based on the field of view and the furthest observation distance as the radius, ensuring the model covers the entire observation range. This fan-shaped region is divided into a first region and a second region, clearly distinguishing areas closer to and farther from the visual sensor, which helps to concentrate resources and attention. A comprehensive value is obtained based on computational complexity, data volume, and computational resource requirements to ensure the model's efficiency and feasibility. The boundary between the first and second regions is determined using an SVM classification model, improving the accuracy and reliability of the division. The area of ​​inefficient regions within the second region that do not meet the requirements of real-time battlefields is calculated using a formula, ensuring the model can accurately identify and process critical areas. Through clear regional division and scientific criteria, data can be managed and analyzed more effectively, improving the overall performance and accuracy of the system.

[0072] First, a comprehensive value is obtained based on computational complexity, computational data volume, and computational resource requirements. Then, the boundary between the first and second regions is determined using an SVM classification model. Finally, the area of ​​each inefficient region in the second region that does not meet the requirements of real-time battlefield is calculated using a formula. The comprehensive value calculation formula is as follows:

[0073]

[0074] in, This is a composite value. Weights for calculating complexity, To calculate the weight of resources, To calculate the weight of the data volume, To calculate the normalized value of the complexity, To calculate the normalized value of the resource, To calculate the normalized value of the data volume;

[0075] To explain further, The overall value represents the overall computational burden of a certain region. The weight of computational complexity reflects its importance in the overall value. The weight of computing resources reflects their importance in the overall value. The weight of the calculated data volume reflects its importance in the overall value. : The normalized value of computational complexity, representing the computational complexity of this region. : Normalized value of computing resources, representing the computing resource requirements of this region. The normalized value of the calculated data volume represents the amount of data in the region. The specific values ​​of the weights can be determined through experiments or expert experience to ensure the accuracy and operability of the calculation.

[0076] The formula for calculating the total area of ​​the second region is as follows:

[0077]

[0078] in, This represents the total area of ​​the second region. The field of view is in radians. The furthest observation distance, The radius of the first region;

[0079] The formula for calculating the area of ​​each inefficient area in the second region is as follows:

[0080]

[0081] in, For the ( , The area of ​​) inefficient areas, The angular increment for the area of ​​each inefficient region, in radians. The radius increment for the area of ​​each inefficient region. The radius of the first region is... This is a radius index for the area of ​​inefficient regions. An angular index for the area of ​​inefficient regions;

[0082] The normalization calculation formula is as follows:

[0083]

[0084] in, The original value of a certain feature. This is the minimum value of the feature. The maximum value of this feature. The value is the normalized value, ranging from [0, 1].

[0085] The core calculation formula of the SVM classification model is as follows:

[0086]

[0087] in, For the weight vector, For the input feature vector, This is a bias term.

[0088] To explain further, Let be the weight vector, representing the direction of the hyperplane. The input feature vector represents the features of the sample. This is the bias term, representing the intercept of the hyperplane.

[0089] The battlefield situation security level is classified according to battlefield situation prediction factors. The visualization content is determined based on the security level classification results, including: obtaining the area of ​​each inefficient area in the second region; constructing a buffer region at the boundary between the first and second regions, with the two endpoints of the buffer region located on the boundary lines of the first and second regions respectively; constructing an angle from the two endpoints of the buffer region to the center line between the two inefficient areas to form an angled region; determining the curvature of the vertex position of the angled region, using the change in curvature as the data source for prediction factors, and using the prediction factors to classify the battlefield situation security level; the buffer region is defined as a dynamic area set at the boundary between the first and second regions to verify and confirm the reliability of the panoramic visual data.

[0090] The specific division of the region is based on: obtaining the starting point of the area of ​​each inefficient region in the second region, connecting the two ends of the buffer region to the starting points of the two inefficient regions respectively, forming two lines; if the intersection of the two lines is located at the farthest end of the second region, then the region where the intersection is located is determined as the angle construction region.

[0091] Further explanation involves explicitly stating the need to obtain the area of ​​each inefficient region within the second region to ensure data integrity and accuracy. Constructing a buffer zone at the boundary between the first and second regions helps smooth the transition and reduce misjudgments. Establishing an angle between the two inefficient regions using the two endpoints of the buffer zone helps to more accurately identify key areas by obtaining the area where the angle is formed. Using curvature changes as a data source for predictive factors helps to capture subtle changes in terrain and environment. Utilizing predictive factors to classify battlefield situational safety levels improves the accuracy and reliability of classification. It is also explicitly stated that the starting point for obtaining the area of ​​each inefficient region within the second region is needed to ensure data accuracy and completeness. Connecting the two ends of the buffer zone to the starting points of the two inefficient regions, if the intersection of these lines is at the farthest point of the second region, this is identified as the area where the angle is formed. Calculating the curvature of the region's vertices ensures the scientific validity and accuracy of the division.

[0092] Further explanation: The first buffer area is defined as follows: A panoramic visual reliability buffer area is a dynamic region set at the boundary between the first and second areas. It is used to verify and confirm the reliability of the panoramic visual data, ensuring data accuracy and security. The second buffer area is defined as: A route optimization buffer area is a dynamic region set around the travel path based on the known reliability of the panoramic visual data. It is used to adjust and optimize the travel strategy, ensuring the optimal travel route and safety. Relationship explanation: Dependency relationship: First buffer area: Through the preprocessing and verification of the panoramic visual data, the reliability and accuracy of the data are ensured; it serves as the intersection area of ​​the first and second areas. Second buffer area: After verifying the reliability of the panoramic visual data in the first buffer area, the route optimization buffer area is constructed based on this reliable data to optimize the travel route and ensure the optimal travel strategy. Logical order: Step 1: Construct the panoramic visual reliability buffer area to verify the reliability of the panoramic visual data. Step 2: Based on the verification results of the first buffer area, construct the route optimization buffer area to optimize the travel route.

[0093] Determine the curvature at this location, including the following formula for calculating curvature:

[0094]

[0095] in, Let f(x) be the curvature, and f(x) be a function of the curve. Let f(x) be the first derivative. It is the second derivative of f(x).

[0096] To further explain, the differentiation method is to differentiate the function using existing analytical methods, numerical methods, or other methods.

[0097] The construction of predictive factors to classify battlefield situational safety levels includes: recording the curvature calculation values ​​at each location and recording the curvature changes at each location over time, with a recording time interval of t seconds; extracting the curvature changes over time for the forward, side, and rear views from the recorded data; using the curvature changes over time in each orientation as a predictive factor sequence; using an LSTM model with the predictive factor sequence as input to predict the optimal route for the next movement; using the predictive factor sequence to generate a recommended route; using the difference between the optimal route and the recommended route as the criterion for safety level evaluation; and using an RNN model with the difference value as input to evaluate the safety level.

[0098] The method of using the difference between the optimal route and the recommended route as the criterion for judging the safety level includes: acquiring second observation angle data to supplement and verify the difference between the optimal route and the recommended route; if traveling along the optimal route, verifying the rationality of the optimal route using the second observation angle data; if the second observation angle data meets the standard of the optimal route, increasing the recommended position value of the recommended route; if the second observation angle data does not meet the standard of the optimal route, decreasing the recommended position value of the recommended route.

[0099] Further explanation involves recording the calculated curvature values ​​at each location to ensure data integrity and accuracy. Simultaneously, the curvature changes at each location over time are recorded to capture dynamic changes in terrain and environment. Forward, side, and rearward views: The curvature changes over time in the forward, side, and rearward view areas are obtained to ensure data diversity and representativeness. LSTM model: The LSTM model is used to predict the optimal route for the next step, improving the accuracy and reliability of the prediction. RNN model: The RNN model is used to evaluate the safety level, ensuring a more scientific and accurate assessment. The difference between the optimal route and the recommended route is used as the criterion for safety level evaluation, ensuring the comprehensiveness and objectivity of the assessment.

[0100] The above steps form a closed-loop system. By continuously collecting new data and feedback, the system is optimized, including: obtaining the difference between the optimal route and the recommended route, and using the difference as the scaling range in the range scaling model, i.e., the buffer area; the buffer area is defined as a dynamic area set around the travel path to adjust and optimize the travel strategy.

[0101] If the difference between the optimal route and the recommended route is negative, the buffer zone moves closer to the first region by shifting its boundary k meters towards the first region. If the difference is positive, the buffer zone moves closer to the second region by shifting its boundary k meters towards the second region. If the difference is zero, the second region expands outward according to the Transformer model by widening its boundary. These steps form a closed-loop system, which is then continuously optimized by collecting new data and feedback.

[0102] Further explanation: The difference between the optimal route and the recommended route can be calculated using Euclidean distance. Obtaining the difference between the optimal and recommended routes: Purpose: To assess the gap between the current recommended route and the ideal route to determine the range requiring adjustment. Method: Calculate the difference between the optimal and recommended routes. By quantifying the difference, the buffer zone can be adjusted more precisely, improving the system's adaptability and accuracy. The difference result is used as the scaling range in the range scaling model, i.e., the buffer zone: Definition: The buffer zone is a dynamic area set around the travel path for adjusting and optimizing the travel strategy. Function: The buffer zone dynamically adjusts according to the difference in the travel route, ensuring the flexibility and safety of the travel strategy. Dynamically adjusting the buffer zone effectively addresses travel needs under different environments, improving the system's robustness and adaptability. Buffer zone adjustment: Negative value case: Condition: The difference between the optimal and recommended routes is negative. Operation: Move the boundary of the buffer zone k meters towards the first region. When the recommended route is better than the optimal route, shrink the buffer zone to reduce unnecessary resource consumption and improve travel efficiency. Positive Value Case: Condition: The difference between the optimal travel route and the recommended travel route is positive. Operation: Move the boundary of the buffer zone k meters towards the second zone. When the recommended route is worse than the optimal route, expand the buffer zone to increase the safety margin and improve the safety of travel. Zero Value Case: Condition: The difference between the optimal travel route and the recommended travel route is 0. Operation: Expand the boundary of the second zone outward according to the Transformer model. When the recommended route is consistent with the optimal route, expand the second zone to cover more potential risk areas, improving the system's predictability and response capability. Closed-Loop System Operation Mechanism: Data Collection: Continuously collect new data, including real-time data of the travel route and environmental data. Feedback Mechanism: Provide feedback to the system based on the collected data and actual travel effects. Optimization Process: Use feedback data to adjust model parameters and optimize the travel strategy and buffer zone settings. Through the closed-loop system, the system continuously learns and improves, enhancing the accuracy and safety of the travel route and ensuring the long-term stability and reliability of the system.

[0103] Example 2: Based on Example 1, a battlefield situation awareness management system based on panoramic vision includes the following:

[0104] Data acquisition and preprocessing module: used to acquire panoramic visual data and perform preprocessing;

[0105] Field of view classification module: used to classify the field of view;

[0106] Range scaling model module: Used to build range scaling models and obtain battlefield situation prediction factors;

[0107] Security Level Classification and Visualization Module: Used to classify the security level of the battlefield situation based on battlefield situation prediction factors and determine the content to be displayed in the visualization.

[0108] Feedback optimization module: Used to optimize the system by continuously collecting new data and feedback;

[0109] Communication and storage module: responsible for data transmission and storage.

[0110] The present application has been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present application. Therefore, the content of this specification should not be construed as a limitation of the present application.

Claims

1. A battlefield situation awareness management method based on panoramic vision, characterized in that, The method comprises the following steps: S1: acquiring panoramic visual data and preprocessing, and then classifying the field of view range; S2: constructing a range scaling model according to the field of view range classification result, obtaining a battlefield situation prediction factor by using the range scaling model; constructing a scaling range model using the front view area, rear view area and side view area data of the first observation angle; acquiring a training set of the front view area, rear view area and side view area data, inputting the training set data into the scaling range model to obtain a trained scaling range model, constructing a sector area with the field of view range as the benchmark and the farthest observation distance as the radius; dividing the sector area into a first area and a second area, wherein the side close to the visual sensor is the first area and the side away from the visual sensor is the second area; The specific division basis is: first, obtaining a comprehensive value according to the calculation complexity, calculation data volume and calculation resource demand, then determining the division line of the first area and the second area in combination with the SVM classification model, and finally obtaining the area of each low-efficiency area in the second area that does not meet the real-time battlefield requirements through formula calculation, the comprehensive value calculation formula is as follows, wherein, is a comprehensive value, is a weight of the computational complexity, is a weight of the computational resource, is a weight of the computational data amount, is a normalized value of the computational complexity, is a normalized value of the computational resource, is a normalized value of the computational data amount; The total area calculation formula of the second area is as follows, wherein, is the total area of the second region, is the field of view range in radians, is the farthest observation distance, is the radius of the first region; The area calculation formula of each low-efficiency area in the second area is as follows, wherein, is the (i)th low-efficiency area area, , )th low-efficiency area area, is the angle increment of each low-efficiency area area, in radian, is the radius increment of each low-efficiency area area, is the radius of the first area, is the radius index of the low-efficiency area area, is the angle index of the low-efficiency area area, and the battlefield situation prediction factor is obtained by using the trained stretching range model. S3: classifying the battlefield situation security level according to the battlefield situation prediction factor, and determining the visual display content according to the security level classification result; S4: forming a closed loop system by S1, S2 and S3, and optimizing the system by continuously collecting new data and feedback.

2. The battlefield situation awareness management method based on panoramic vision of claim 1, wherein, The panoramic visual field range data is acquired and preprocessed, and then the field of view range is classified, which comprises: the preprocessing comprises data cleaning and standardization processing; the classification of the field of view range comprises: dividing the field of view range into a front view area, a rear view area, a side view area, an upper view area and a bottom view area; The angle of the front view area, the rear view area, the side view area and the upper view area is the first observation angle, and the angle of the bottom view area is the second observation angle.

3. The battlefield situation awareness management method based on panoramic vision of claim 1, wherein, The first observation angle, the rear view area, the side view area, the upper view area and the bottom view area are as follows: The normalization calculation formula is as follows, where, is the original value of a certain feature, is the minimum value of the feature, is the maximum value of the feature, is the normalized value, ranging from [0, 1]; The core calculation formula of the SVM classification model is as follows, wherein, is a weight vector, is an input feature vector, is a bias term.

4. The battlefield situation awareness management method based on panoramic vision of claim 1, wherein, The battlefield situation security level is classified according to the battlefield situation prediction factor, and the visual display content is determined according to the security level classification result, which comprises: acquiring the area of each low-efficiency area in the second area, constructing a buffer area at the junction of the first area and the second area, and the two end points of the buffer area are located on the boundary lines of the first area and the second area respectively; constructing an included angle between the two end points of the buffer area to the center line between the two low-efficiency areas to form an included angle area; determining the curvature of the vertex position of the included angle area, taking the change of the curvature as the data source of the prediction factor, and classifying the battlefield situation security level by using the prediction factor; the buffer area is defined as a dynamic area at the junction of the first area and the second area, which is used to verify and confirm the reliability of the panoramic visual data; The specific division of the region is based on: obtaining the starting point of each low-efficiency area in the second region, connecting the two ends of the buffer region with the starting points of the two low-efficiency areas respectively to form two connecting lines; if the intersection of the two connecting lines is located at the farthest end of the second region, the region where the intersection is located is determined as the angle construction region.

5. The battlefield situation awareness management method based on panoramic vision of claim 4, wherein, The curvature of the position is determined, including: the calculation formula of the curvature is as follows, where, is the curvature, f(x) is a function of the curve; is the first derivative of f(x), is the second derivative of f(x).

6. The battlefield situation awareness management method based on panoramic vision of claim 4, wherein, The prediction factor is used to classify the battlefield situation safety level, including: recording the curvature calculation value of each position, and recording the change of the curvature of each position with time, the time interval of recording is t seconds; extracting the change of the curvature of the front view area, the side view area and the rear view area with time from the recorded data; taking the change of the curvature in each direction with time as a prediction factor sequence; using the LSTM model to take the prediction factor sequence as input to predict the best route for the next step; using the prediction factor sequence to generate a recommended route for the march; taking the difference between the best route for the march and the recommended route for the march as the judgment standard of the safety level; using the RNN model to take the difference value as input to judge the safety level; Wherein, the difference between the best route for the march and the recommended route for the march is taken as the judgment standard of the safety level, including: obtaining second observation angle data for supplementing and verifying the difference between the best route for the march and the recommended route for the march; if the best route for the march is used, the second observation angle data is used to verify the rationality of the best route for the march; if the second observation angle data meets the standard of the best route for the march, the recommended position value of the recommended route for the march is increased; if the second observation angle data does not meet the standard of the best route for the march, the recommended position value of the recommended route for the march is reduced.

7. The battlefield situation awareness management method based on panoramic vision of claim 1, wherein, The S1, S2 and S3 steps form a closed loop system, which is optimized by continuously collecting new data and feedback, including: obtaining the difference between the best route for the march and the recommended route for the march, taking the difference value as the expansion range in the range expansion model, that is, the buffer region; the definition of the buffer region is a dynamic region set around the march path, which is used to adjust and optimize the march strategy; If the difference between the best route for the march and the recommended route for the march is negative, the buffer region is close to the first region, and the specific operation is to move the boundary of the buffer region to the first region by k meters; if the difference between the best route for the march and the recommended route for the march is positive, the buffer region is close to the second region, and the specific operation is to move the boundary of the buffer region to the second region by k meters; if the difference between the best route for the march and the recommended route for the march is 0, the second region is expanded outward according to the Transformer model, and the specific operation is to expand the boundary of the second region; then the S1, S2 and S3 steps form a closed loop system, which is optimized by continuously collecting new data and feedback, to ensure the continuous improvement and performance improvement of the system.

8. A management system for battlefield situation awareness based on panoramic vision according to any one of claims 1-7, characterized in that, It includes: Data acquisition and preprocessing module: used for acquiring panoramic vision data and preprocessing; Field of view range classification module: used for classifying the field of view range; Range expansion model module: used for constructing a range expansion model to obtain battlefield situation prediction factors; The security level classification and visualization module is used for classifying the security level of the battlefield situation according to the battlefield situation prediction factors and determining the content of the visual display; The feedback optimization module is used for optimizing the system by continuously collecting new data and feedback; The communication and storage module is used for data transmission and storage.

Images